Biochem. J. (1993) 295, 227-232 (Printed in Great Britain) 227

Identification of hepatocyte nuclear factor-3 binding sites in the Clara cellsecretory protein geneColin D. BINGLE and Jonathan D. GITLIN*Edward Mallinckrodt Department of Pediatrics, Washington University School of Medicine, St. Louis, MO 63110, U.S.A.

To determine the mechanisms of cell-specific gene expression inthe developing pulmonary epithelium the Clara cell secretoryprotein (CCSP) gene promoter was analysed by DNAase Ifootprinting. A prominent site of protein-DNA interaction wasdetected from nucleotides - 132 to -76 using nuclear extractfrom mouse lung and human H441 cells. Mobility shift analysisrevealed that an oligonucleotide corresponding to this regioninteracted with multiple proteins from lung and H441 cell nuclearextracts. Analysis of the nucleotide sequence of this regionidentified two potential binding sites for hepatocyte nuclearfactor 3 (HNF-3), and consistent with this finding binding to thisCCSP oligonucleotide was specifically competed for by anoligonucleotide corresponding to the HNF-3-binding site from

INTRODUCTION

The pulmonary epithelium originates from an outpouching ofembryonic gut endoderm into the surrounding fetal mesenchyme.Inductive signals mediated by cell-cell contact and secretionfrom this mesenchyme result in the formation of a contiguousepithelium from the trachea to the alveolus [1,2]. Geographicdifferentiation along this epithelium results in the developmentof specialized epithelial cells subserving the unique functions ofthe respiratory system. In the bronchiolar epithelium, suchdifferentiation results in the formation of non-ciliated epithelial(Clara) cells which synthesize proteins essential for surfactantfunction and pulmonary host defence [3,4]. This epithelium is thesite of expression of the cystic fibrosis transmembrane con-ductance regulator protein during human fetal development andis a common site of anaplastic transformation [5,6]. Despite theobvious importance of the bronchiolar epithelium in lung de-velopment and function little is known about the mechanisms ofgrowth and differentiation in this tissue.

Clara cell secretory protein (CCSP) is an abundant 10 kDaprotein present in lung secretions and shown by immuno-histochemistry to localize to secretory vesicles within the bron-chiolar epithelium [7]. Although the biological function of CCSPis unknown, the amino acid sequence of this protein is highlysimilar to rabbit uteroglobin and identical to a polychlorinated-biphenyl-binding protein isolated from rat lung [8]. The geneencoding CCSP has been isolated in the rat and studies usingRNA-blot analysis indicate that CCSP gene expression is con-

fined to the lung [8,9]. In situ hybridization studies reveal thatwithin the lung parenchyma CCSP gene expression is cell-specific,being confined to Clara cells of the terminal and respiratorybronchioles [10,11].CCSP gene expression is an informative developmental marker

within the bronchiolar epithelium recapitulating the cellular

the mouse transthyretin gene. Mobility shift of the CCSPoligonucleotide was supershifted using antisera specific to HNF-3ac and HNF-3,/, and HNF-3a and HNF-3,8 translated in vitrowere found to bind specifically to this same oligonucleotide. Co-transfection of HNF-3a- and HNF-3/?-expression plasmids in-creased cell-specific reporter gene activity in H441 cells trans-fected with a CCSP-CAT gene chimeric construct containingthis - 132 to -76 region. Taken together, these results suggest arole for HNF-3 in mediating cell-specific CCSP gene expressionwithin the bronchiolar epithelium. These findings support thehypothesis that members of the HNF-3 'forkhead' family oftranscription factors determine gene expression and cell fate inmultiple cell lineages derived from the primitive gut endoderm.

differentiation in the distal respiratory epithelium during fetaland post-natal development [11]. Studies in transgenic mice haverevealed that cis-acting elements within 2.25 kb of the 5' flankingregion of the rat CCSP gene are sufficient to direct tissue- andcell-specific expression of a chimeric CCSP-reporter gene [12,13].Characterization of the cis-acting elements determining cell-specific gene expression can be a valuable approach to definingthe mechanisms of cellular differentiation in tissues where lineageanalysis is not available [14]. In this present study we havecharacterized the 5' flanking region of the rat CCSP gene toelucidate the cis- and trans-acting factors responsible for cell-specific gene expression within the bronchiolar epithelium.

MATERIALS AND METHODSPreparation of nuclear extractsNuclei were isolated from organs of adult mice as described [15].All procedures were performed at 4 'C. Isolated nuclei were lysedin a solution containing 10 mM Hepes, pH 7.9/100 mM KCl/3 mM MgCl2/0. 1 mM EGTA/1 mM dithiothreitol (DTT)/10%(v/v) glycerol and the proteins were precipitated with(NH4)2SO4. The protein pellet was resuspended in 25 mM Hepes,pH 7.6/40 mM KCl/0.1 mM EDTA/1 mM DTT/10%0 glyceroland dialysed overnight against two changes of the same buffer.Precipitated proteins were cleared from the solution by spinningin a microfuge for 10 min. Protein concentration was determinedfrom the supernatants using the Bradford assay [16].

DNAase I footprintingIndividual fragments of the 5' flanking region of the CCSP genewere prepared by restriction endonuclease digestion of genomicclones encoding defined regions of the rat CCSP gene [11]. Thegenomic clones corresponding to this region were isolated and

Abbreviations used: CCSP, Clara cell secretory protein; HNF-3, hepatocyte nuclear factor 3; DTT, dithiothreitol.* To whom correspondence should be addressed at: St. Louis Children's Hospital, One Children's Place, St. Louis, MO 63110, U.S.A.

Biochem. J. (1993) 295, 227-232 (Printed in Great Britain) 227

228 C. D. Bingle and J. D. Gitlin

characterized from a rat genomic library as described [11]. DNAfragments were isolated on DEAE paper, treated with calf-intestine alkaline phosphatase, end-labelled with [y-32P]ATPand T4 polynucleotide kinase, digested with restriction endo-nucleases, and recovered by salt elution after electrophoresisthrough 20% (w/v) polyacrylamide gels [17]. DNAase I reactionswere performed using 1-2 ng of labelled DNA (specific radio-activity approx. 108 c.p.m./jug) incubated in 50 1l of 25 mMTris/HCl, pH 7.9/6.25 mM MgC12/0.5 mM EDTA/5OmMKCl/0.5 mM DTT/10% (v/v) glycerol/i jug poly (dI-dC) at4 °C for 15 min in the presence or absence of nuclear extracts asindicated [18]. A predetermined amount of DNAase I was addedto the reaction for 1 min and the reaction was then stopped bythe addition of 100 jul of 20 mM NaCl/20 mM EDTA/1 %SDS/250 ,ug/ml tRNA. DNA was phenol/chloroform-extracted,precipitated in ethanol and electrophoresed through 8 % (w/v)acrylamide-urea gels. Nucleotide sequence reactions ofthe regionbeing footprinted were run in parallel in these same gels foraccurate identification of the size and nucleotide sequence offootprinted regions. After electrophoresis gels were fixed, driedand exposed to film at -80 'C.

Gel-mobility-shiff assaysOverlapping complementary oligonucleotides were synthesizedon an Applied Biosystems oligonucleotide synthesizer for use asprobes in the mobility-shift assays. The complementary oligo-nucleotides were annealed and labelled by filling in the over-lapping regions with [32P]dNTPs and Klenow polymerase. Eachbinding reaction used 1 ng of probe (specific radioactivity107 c.p.m./jug). Binding reactions were performed at 20 °Cin a 20 jul reaction mix containing 20 mM Hepes, pH 7.6,40 mM KCI, 2 mM MgCl2, 1 mM DTT, 0.5 mM EGTA, 4%(w/v) Ficoll, 1 jug of poly(dI-dC) and nuclear protein (1-10 ug)[19]. The probe was added last and the reaction was allowed toproceed for 30 min. At the end of the binding reaction sampleswere electrophoresed through 60% (w/v) polyacrylamide gels[30: 1, acrylamide/bis, 0.25 x Tris-borate-EDTA buffer (TBE)]at 20 'C. Gels were dried without fixing and exposed to film at-80 'C. In some reactions unlabelled competitor oligonucleo-tides were included at the concentration outlined in the text.Oligonucleotides used in these studies corresponded to sequencesfor the mouse transthyretin gene hepatocyte nuclear factor-3(HNF-3) site [19] (5'-TGACTAAGTCAATAATCACAATC-AGCAGGT-3') and the human immunoglobulin-light-chaingene NF-KB site [20] (5'-ACAGAGGGGACTTTCCGTCTCC-3'). For antibody supershift experiments 1 jul of preimmuneserum or 1 jul of antiserum specific to HNF-3 [21] or C/EBP [22]family members was added to the reaction and incubated for15 min at 20 'C before the addition of the labelled probe. In allcases titration studies were performed using dilutions of thesespecific antibodies to be certain that the specificity of interactionwas maintained.

in the mobility-shift assay as outlined above, except that theamount of poly(dI-dC) was increased to 2.5 jug. Two sheets ofX-ray film were used to produce the autoradiographs with thefilm in proximity to the gel blocking the 35S signal from thetranslated protein.

Plasmids and construction of chimeric genesFull-length cDNAs corresponding to rat HNF-3a and HNF-3,inserted in either pBluescript or a CMV expression plasmid wereused for translation in vitro and transfection respectively [21].Cells were transfected as 70% confluent monolayers using 2.5 jugof CCSP-CAT constructs and an equal amount of carrier DNA(pBluescript) and 20 jul of Lipofectin. Lipofectin/DNA mixtureswere left on cells for 5 h, after which time the cells were washedthree times with PBS and re-fed with complete growth medium[24]. In some experiments carrier DNA was replaced withexpression plasmids for HNF-3a or HNF-3,/. To control forefficiency of transfection 1 jug of pSV-,#gal was co-transfectedinto cells during all transfections. Reporter gene assays wereperformed as previously described [17].

RESULTSDNAase I footprinting of the CCSP promoterTo identify potential cis-acting elements determining cell-specificCCSP gene expression the 5' flanking region of the CCSP genewas analysed by DNAase I footprinting. Footprinting studiesusing nuclear extracts from mouse lung identified four prominentfootprint regions using DNA fragments containing nucleotides-274 to + 37 and - 1106 to -658 from the rat CCSP gene(Figure 1, lanes 1-8). Additional footprints were not observed inthe remaining 5' flanking sequence up to and including -2072.

Nuclearextract - - -J

-132

1* § l -76

-. -29

-22

S +1

123 4

CD 0)c c

j; s__-132

-.-:- -76

9 10 11 12

_ _,,

Ie .

sw.i2..... w

s ,'w w; ::

8D |F.::.llF ....

wi *.

; * * *_ _

:.':

ieL ::.s . :6

:. E::

* _ e:

:.7i.

* ::.

5 6 7 8

Translation in vitro and gel-mobility-shift assays

Linearized plasmids containing full-length HNF-3a or HNF-3,fcDNA sequences were used as templates for transcription usingT7 RNA polymerase [23]. Translation in vitro was performedusing nuclease-treated rabbit reticulocyte lysate with [35S]methi-onine and reaction products were analysed by SDS/PAGE toconfirm the size of the translated product. Reticulocyte lysate(2.5 jul), containing the translated protein, was then used directly

Figure 1 DNAase I footprinting of the rat CCSP promoter

DNAase footprinting was performed as described in the text in the absence (lanes 1, 2, 5,6, 9 and 10) or presence of 50 ug (lanes 3 and 7) or 100 ug (lanes 4, 8 and 11) of mouselung nuclear extract, and 100 4g of H441 -cell nuclear extract (lane 12) using end-labelled DNAprobes corresponding to nucleotides -313 to +37 (lanes 1-4 and 9-12) or -1143 to-886 of the 5' flanking region of the rat CCSP gene, labelled on the non-coding strand. AfterDNAase digestion the reaction products were resolved on an 8% (w/v) acrylamide sequencinggel, fixed, dried and analysed by autoradiography. The locations of the regions protected fromDNAase digestion are indicated at the side of the figure. The -132 to -76 region is referredto as CCSP region 1.

Clara cell gene expression

-132 -HNF-3-- -76

(a) 5' GAAAAGAGATTATTTQCTTATTGCATGGAGATGACTAAGTAAATAGTGCAATTTCTT 3'

3' CTTTTCTCTAATAAAC''ATAACGTACCTCTACTQATTCATTTATCACGPTAAAGAA 5'

-HNF-3-

(bI 5' GAAAAGAGATTATTTGCT.ATTGCATGGAGAT 3' (WT)

5' GAAAAGAGATGC0TQCTTMTTGCATGGAGAT 3' (MU)

(C) (WT) 5' AGATGACTAAGTAMTAGTGCAATTTCTT 3'

(MU) 5' AGATGACTA?GTACCGCGTGCAATTTCTT 3'

(dl Rat CCSP -122 to-1 12 TATTTGCTTAT

-88 to -98 TATTTACTTAG

(a)

CompetitorNuclear extract

0L Ur)LLzI

(L EL cn?'

u

U)Le

C) Lz

+ + + + + +

a-

b-

...

-8:'.

Mouse transthyretin

Mouse a,i-antitrypsin

-1 06 to-96 TATTGACTTAG-130 to-140 TATTITGTGTAG

-376 to -366 TATGACTTTG

-195 to-185 CATTGATTTAG

Human apolipoprotein B +863 to +873 CATTTGCTTAT

Rabbit uteroglobin -131 to -121 TATTTACTTAT-94 to-104 TAtTTACTTGG

Figure 2 Nucleotide sequence of the CCSP region I

The sequence of the CCSP region (a) showing potential binding sites for HNF-3 (highlighted)as well as the sequences of the shorter 5'CCSP (b) and 3' CCSP (c) oligonucleotides usedin the gel-mobility-shift studies are shown. Both wild-type (WT) and mutated (MU) oligonucleotidesequences are shown. (d) The sequences represent a comparison of the putative HNF-3-bindingsites from the rat CCSP gene with known HNF-3-binding sites as well as recently identifiedsequences in the rabbit uteroglobulin promoter. Highlighted nucleotides represent the mostconserved bases.

1 2 3 4

v':: .. -:'.

5 6 7

(b)32p probe

CompetitorNuclear extract

13 :CCSP 3'

ZI in zn zo

o- E E C

r - - c) u ut - + + + + +

c- .....W...

d iii; U.

CCSP3'M

?n ?n_ O U)

+ + +

_

A prominent site of protein-DNA interaction was detected from-132 to -76 (CCSP region I), which also footprinted usingnuclear extract from H441 cells (Figure 1, lanes 9-12). Wefocused initially on this region because nucleotide sequenceanalysis revealed two potential sites for the liver-specific tran-scription factor HNF-3 arranged in opposite directions along theDNA. Comparison of these regions with HNF-3 sites shownpreviously to bind HNF-3 in vitro revealed a high degree ofsimilarity, consistent with a derived consensus sequence (Figure2).

Interaction of CCSP I site oligonucleotides with nuclear proteinsA double-stranded oligonucleotide corresponding to CCSP I(Figure 2a) was end-labelled with [y-32P]ATP and used as a probein gel-mobility-shift reactions with nuclear extracts from H441cells. This reaction gave rise to a number of specific retardedcomplexes which were competed with a 100-fold molar excess ofunlabelled oligonucleotide (Figure 3a, lanes 2 and 3). As with thefootprint studies similar results were obtained using mouse lungnuclear extract (results not shown); thus for the remainder ofthese studies H441-cell extracts were used. Competition was alsoperformed using oligonucleotides corresponding to known bind-ing sites for HNF-3 and NF-KB. Each retarded complex was

specifically inhibited by 100-fold molar excesses of unlabelledCCSP I and HNF-3 oligonucleotides (Figure 3a, lanes 3 and 4).Competition for all bands was also achieved using the 3' and 5'CCSP I wild-type oligonucleotides shown in Figure 2 (Figure 3a,lanes 5 and 6) which contain a single HNF-3-binding site. Theseeffects were specific because no competition was observed withan oligonucleotide corresponding to the NF-KB binding site(Figure 3a, lane 7). Gel-mobility-shift assays performed using a

1 2 3 4 5 6 7 8 9 10

Figure 3 Gel mobiiity shift of H441-cell nuclear extracts with 32P-labelledCCSP I oligonucleoUdes

(a) Gel-mobility-shift reactions were performed with 5 ,ug of H441 -cell nuclear extract and 32p-labelled CCSP oligonucleotide as a probe alone (lane 1), in the absence of competitoroligonucleotide (lane 2), or in the presence of 1 00-fold molar excess of unlabelled specific (lanes3-6) and non-specific competitors (lane 7) including CCSP oligonucleotides containing a singleputative HNF-3-binding site. Reactions were resolved in 6% 0.25 x TBE gels as outlined in theMaterials and methods section. (b) Gel-mobility-shift reactions were performed as outlined using5 jug of H441-cell nuclear extract and 32P-labelled CCSP 3' oligonucleotide alone (lane 1),in the absence (lane 2) or presence of 100-fold molar excess unlabelled 3' wild-type (lane 3),3' mutated (lane 4), 5' wild-type (lane 5), or 5' mutated (lane 6) oligonucleotide. Also shownis the 32P-labelled CCSP 3' mutated oligonucleotide alone (lane 7), in the absence (lane 8)or presence of 100-fold molar excess unlabelled 3' mutated (lane 9), or wild-type (lane 10)oligonucleotide.

32P-labelled HNF-3 oligonucleotide resulted in the formation ofmultiple complexes using H441-cell nuclear extract, each ofwhich was specifically competed with by the CCSP I oligo-nucleotides (results not shown).To examine further the nature of the individual HNF-3 sites in

the CCSP promoter 3' oligonucleotide was 32P-labelled and usedas a probe in mobility-shift assays (Figure 3b). Four distinctbands were observed using the 3' oligonucleotide (Figure 3b,lane 2) and each was competed for by a 100-fold molar excess

of 3' and 5' CCSP I oligonucleotides (Figure 3b, lanes 3 and 5). Incontrast only some of the complexes were competed for using 5'and 3' oligonucleotides in which the putative HNF-3-binding site

229

230 C. D. Bingle and J. D. Gitlin

had been mutated as shown in Figure 2 (Figure 3b, lanes 4 and6). Consistent with this observation, when mobility shift wasperformed using the 32P-labelled 3' mutated oligonucleotide, onlythose protein-DNA complexes not competed for in the abovestudies were observed (Figure 3b, lanes 7, 8 and 9 versus lanes 4and 6). Identical results were obtained using the CCSP I 5' wild-type and mutated oligonucleotides as probes (results not shown).

HNF-3a and HNF-3fp bind to CCSP I

The gel-mobility-shift experiments suggested that an HNF-3-likeprotein present in H441-cell nuclear extracts was capable ofbinding to the CCSP I site. To examine this in greater detailspecific antisera to HNF-3a and HNF-3,8 were used in gel-mobility-shift experiments (Figure 4a). As seen previously theCCSP I oligonucleotide retarded four specific complexes fromH441-cell nuclear extracts, each of which were competed for bya 100-fold molar excess of the specific CCSP I oligonucleotidebut not an equivalent excess of an oligonucleotide correspondingto NF-KB (Figure 4a, lanes 1-4). When antisera were added tothese reactions, a supershift of the CCSP I complexes wasobserved with antiserum to HNF-3a but not antiserum to HNF-3,/ or C/EBPa (Figure 4a, lanes 6 versus lanes 8 and 10).Although a change in gel-shift pattern was observed with thepreimmune sera this effect was non-specific, being present witheach added antiserum (Figure 4a, lanes 5-10). When gel-mobility-shift analysis was performed using the 3' CCSP I oligonucleotidein the presence of preimmune of HNF-3a antisera (Figure 4a,lanes 12 and 13) a similar supershift of some complexes wasobserved. The complexes which supershifted were identical tothose characterized as HNF-3 by gel mobility shift using wild-type and mutant oligonucleotides (Figure 3b).To demonstrate directly that the sequence within the CCSP I

site was capable of binding to HNF-3 proteins, gel-mobility-shiftassays were performed using HNF-3a and HNF-3,8 proteinstranslated in vitro. Each protein was translated from cRNAtranscribed from specific HNF-3 plasmids and the translatedproduct was shown to be of the correct size (a, 50 kDa; /,47 kDa) by SDS/PAGE (results not shown). A retarded complexofidentical size was detected when reticulocyte lysates containingeither HNF-3a or HNF-3,8 were incubated with 32P-labelledoligonucleotides corresponding to the HNF-3 site (Figure 4b,lanes 1-4) or the CCSP I site (Figure 4b, lanes 5-10). In each casethe complex was inhibited by an excess of the specific competitorbut not by an oligonucleotide corresponding to the NF-KB site.An identical specific complex was detected in these studies usingthe CCSP I 3' oligonucleotide and reticulocyte lysate containingHNF-3a (Figure 4b, lane 12) and this was competed by anexcess of wild-type but not mutated 3' oligonucleotide (Figure4b, lanes 13 and 14). The differences in intensity of bound HNF-3 with each oligonucleotide reflected differences in specific activityof the oligonucleotides used as probes. The additional bandsfound with the CCSP I 3' oligonucleotide represent degradationproducts of HNF-3ac following storage of lysate at -20 °C aseach band was eliminated in competition experiments.

Transactivation of a CCSP4AT constructThe data so far suggested that HNF-3a and , interacted invitro with the CCSP I region. To determine if the HNF-3 sitescontained within the CCSP I region were functional, transienttransfection studies were performed using a CCSP-CAT genechimeric construct containing 1.17 kb of the rat CCSP gene 5'

(a)

Antibody ...

32p probe

CompetitorNuclear extract

HNF-3a HNF-3P C/EBPa_+ -_ + 'l - +

CCSP

UU LL

_ -u z - _ _ _ _ __ + + + + + + + + +

.:

_, j-: -lb :: :_r ........... " " :!_ _..1". F- ......... .. ................. _ ... --. l* ._ . _ _ _- 312: a_ _: :.1_ _. . -._ ... . ................................. _ _ ._ ._._. _ _._ _: _ :-: .:_P _._ _.

*=.. .:^s:.3e. |- - -... _; ......... :R.c . _ _ ._ _ .......................... _ _.93ii: 3_ .Sf.§'g;HE'. .iSiEil; _t'_._lE _s.: . . _ . ._. _. _._: _ _, ._',,.','.. .. .:; : . 2 ' :! ' ' : fi1 2 3 4 5 6 7 8 9 10

HNF-3ca11

IS-+3CCSP 3'

_ + + +

.........._

11 12 13 14

(b)

32p probe: HNF-3 CCSP CCSP 3'

Protein: HNF-3a HNF-3P3 HNF-3a

c) Q- CM CL CL Q

Competitor: - - z uL -C-

I Zu

Z )u Zu u

_ ._ .* . z. _: _

im.

1 2 3 4 5 6 7 8 9 10 11 12 13

Figure 4 Gel-mobility-shift assay of H441-cell nuclear extracts andtranslated proteins

(a) Gel-mobility-shift reactions were performed as outlined in the Materials and methods sectionwith 5 ,ug of H441-cell nuclear extract and the 32P-labelled CCSP oligonucleotide alone (lane1), in the absence (lane 2) or presence of 100-fold molar excess unlabelled CCSP (lane 3)or NF-KB (lane 4) competitor oligonucleotides and in the presence of 1 ,ul of either preimmuneserum (lanes 5, 7 and 9) or 1 ,ul of antiserum to HNF-3a (lane 6), HNF-3,8 (lane 8) or C/EBPa(lane 10). Gel-mobility-shift assay was also performed using the 32P-labelled CCSP 3'oligonucleotide alone (lane 11) or with H441-cell nuclear extract alone (lane 12) and in thepresence of either preimmune serum (lane 13) or antiserum to HNF-3a (lane 14). Reactionswere resolved in 6% 0.25 x TBE gels as outlined in the Materials and methods section. (b)Translation reactions in vitrowere performed using cRNAs for HNF-3cc and HNF-3p8as outlinedin the Materials and methods sections. A sample (2.5 ,#I) of reaction products was used toperform gel-mobility-shift assays using either 32P-labelled HNF-3 (lanes 1-4) or CCSP (lanes5-10) oligonucleotides as probes. In lanes 2-7 the translation reaction was primed with HNF-3a RNA and in lanes 8-10 the translation reaction was primed with HNF-3,/ RNA. Reactionswere performed in the absence (lanes 2, 5 and 8) or presence of unlabelled specific (lanes 3,6 and 9) or non-specific (lanes 4, 7 and 10) competitor oligonucleotides. Mobility shift was alsoperformed using 32P-labelled CCSP 3' oligonucleotide wAh in vitro translated HNF-30c in theabsence (lane 11) or presence of wild-type (lane 12) or mutated (lane 13) CCSP 1 3'oligonucleotide. The reactions were resolved in 6% 0.25 x TBE gels as outlined in the Materialsand methods section.

flanking region. As can be seen in Figure 5, the CCSP 1.17 CATplasmid but not a pOCAT-BS construct was transcriptionallyactive in H441 cells (Figure 5, lanes 1-3). When this constructwas co-transfected with expression plasmids for either HNF-3aor HNF-3/8 there was a marked increase in CAT activity inH441-cell extracts (Figure 5, lanes 4 and 5). No increase in CATactivity was observed when these same transfections were re-peated using an equivalent amount of an expression plasmid forthe transcription factor GATA-4 (Figure 5, lane 6). The induction

Clara cell gene expression 231

- H441 cells I r HEP G2 cells sl)m

CV) CY)LL UL

I I (C >0l L)

r- CSP1.17CAT -i 0 0

I?)C' CO) C')

L LL LL LLz z z z

I I I II I

r CCSP1.17CAT- CP-393CAT---

1 2 3 4 5 6 7 8 9 10 11 12 13 14

Figure 5 Co-translection of CCSP-CAT chimeric construct with HNF-3expression plasmids

Transient transfections were performed as outlined in the Materials and methods section ineither H441 (lanes 14) or Hep G2 (lanes 7-14) cells using a CCSP-CAT chimeric constructcontaining 1.17 kb of 5' flanking region of the rat CCSP gene. Cells were transfected with2.5 ,ug of pOCAT-BS (lanes 1 and 7), pSV2CAT-BS (lanes 2, 8), pCCSP1 .17CAT (lanes 3, 9),pCCSP1.17CAT and pCMV-HNF-3a (lanes 4, 10), pCCSP1.17CAT and pCMV-HNF-3/ (lanes5, 11), pCCSP1 .17CAT and pGATA-4 (lane 6), pCP-393CAT (lane 12), pCP-393CAT and pCMV-HNF-3 a (lane 13), pCP-393CAT and pCMV-HNF-3,6 (lane 14). All reactions also contained1 ,ug of pSV,8-Gal as a control, and pBS-SK such that each transfection reaction contained thesame amount of DNA. After incubation for 42 h the cellular extracts were assayed for CATactivity and the reaction products were separated by t.l.c. as described.

of expression seen in H441 cells with the HNF-3 plasmids was

cell specific as a similar effect was not seen when identicaltransfections were performed in Hep G2 cells. In this case theCCSP 1.17 CAT construct was inactive, either alone (Figure 5,lane 9) or when co-transfected with the HNF-3 expressionplasmids (Figure 5, lanes 10 and 11). This lack of induction withHNF-3 was not due to inactivity of the HNF-3 constructs in HepG2 cells because a rat ceruloplasmin gene-CAT constructcontaining HNF-3-binding sites was effectively transactivated inthese cells (Figure 5, lanes 12-14). The effect of HNF-3 on theCCSP 1.17 region was highly reproducible, being consistentlyobserved in six independent transfection assays utilizing differentplasmid preparations. In all experiments variations in trans-fection efficiency were controlled as described in the Materialsand methods section.

DISCUSSIONTransgenic experiments reveal that cis-acting elements within2.25 kb of the 5' flanking region of the CCSP gene were sufficientto direct appropriate lung- and Clara-cell-specific expression ofreporter genes [12,13]. This current study focused on this regionwith DNAase I footprinting identifying several specific footprintsconsistent with transient transfection studies in H441 cells [12].Sequence analysis of the CCSP I region revealed the presence ofputative cis-acting elements for HNF-3 and consistent with thisobservation the mobility gel shift, antibody supershifts and co-

transfection studies presented here all support a role for HNF-3in this region. It is important to note that the CCSP I oligo-nucleotides interacted with multiple proteins in H441-cell nuclearextracts. While it is possible that each complex represents a

different form of HNF-3 it seems most likely that other tran-scription factors also interact with the CCSP I region. Experi-ments using HNF-3-specific antisera and the CCSP oligo-nucleotide, in which the HNF-3-binding sites were mutated,support the latter concept.

Interestingly, two HNF-3 sites are found within CCSP regionI oriented in opposite directions. The orientation of these sitesand their proximity to each other is similar to the arrangementobserved in the HNF-3 sites in the mouse transthyretin genewhich is also transactivated by HNF-3 [19,21]. Although thereare currently no data available to suggest whether one or both ofthese sites is necessary for HNF-3 function, this arrangement isintriguing since an oligonucleotide to each of these regionssuccessfully competed in the gel mobility studies for the HNF-3-retarded complexes. Since no model currently exists for themechanism of HNF-3 binding to DNA it will be important infuture studies to determine whether the bidirectional sitesobserved here for the CCSP gene are essential for HNF-3 activityin vivo. The absence of the HNF-3,f-specific supershift in H441-cell nuclear extract is due to a lack of expression of HNF-3fl inthese cells, as determined by immunoprecipitation (results notshown) as opposed to a specific inhibition of CCSP I-HNF-3flinteraction. The CCSP I oligonucleotide did interact with bothHNF-3a and HNF-3fl in mouse lung extracts and in HNF-3fl-transfected H441 cells (results not shown), but it is not clear fromthe current studies which of these proteins specifically interactswith the CCSP gene promoter in vivo. Previous studies havedemonstrated the presence of HNF-3a and HNF-3,f transcriptsin mouse lung [21] and thus further studies will be required todetermine the role of each HNF-3 member in CCSP geneexpression in vivo.The results shown here clearly demonstrate an increase in

CAT activity after co-transfection with either HNF-3a or -3/3.The cell specificity of these results, as indicated by the Hep G2transfections, suggests that H441 cells contain additional factorsessential for Clara-cell-specific expression of the CCSP gene.Although the H441 cells have many Clara cell characteristics,these cells do not express the endogenous CCSP gene. Never-theless, the H441-cell-specific expression of CCSP gene chimericconstructs observed here is consistent with recent observationsthat endogenous CCSP gene expression is actively repressedearly in lung development and that the cis-acting elementsresponsible for this repression are not contained within 2.25 kbof the 5' flanking region (B. P. Hackett and J. D. Gitlin, un-published work). If such mechanisms are operative in H441 cellsthis may explain the transient expression observed using thisregion, despite the lack of endogenous CCSP gene expression.Consistent with this, transfection of the HNF-3a or HNF-3,/expression plasmids did not activate endogenous CCSP geneexpression in H441 cells.HNF-3 transcription factors are members of the recently

described forkhead family named after the original memberidentified as a Drosophila homeotic mutation [25]. Additionalfamily members are the Drosophila sloppy paired loci [26], anactivin-inducible Xenopus gene XFKH1 [27], two distinct HIV-and interleukin-binding factors from human T-cells [28,29], arodent brain-specific transcription factor BFI [30], a Drosophilaforkhead homologue in Saccharomyces cerevisiae [31], as well asmultiple new family members recently identified in Drosophilaembryo [32]. Each of these factors, including HNF-3a, -3/3 and-3y all have a highly conserved 110 amino acid 'forkhead'domain which has been shown by deletion studies to be a site ofDNA binding as well as two other conserved domains mediatingtranscriptional activation. Many of the forkhead members playa role in determinative events of cell fate during embryogenesis.The studies presented here are the first to demonstrate a role forthis transcription-factor family in the regulation of pulmonaryepithelial cell-specific gene expression. Given the embryologicorigins of the pulmonary epithelium our findings are consistentwith the hypothesis that HNF-3 proteins may be important in

U)

04CL)

U)

ca)

0Q

232 C. D. Bingle and J. D. Gitlin

determining gene expression and cell-type determination inmultiple epithelial cell-types derived from the primitive gutendoderm. Recent studies identifying unique forkhead familymembers specific to the lung also support the concept that thesetranscription factors may play a role in cell type determination inthe developing pulmonary epithelium [33].

We thank A. L. Schwartz and D. Wilson for critical review of the manuscript, J.Darnell for the HNF-3 plasmids and antisera, D. Wilson for the GATA-4 expressionplasmid, S. McKnight for the C/EBP plasmids and antisera, M. Migas for technicalassistance and D. Howard for manuscript preparation. This research was supportedby funds from grant HL41536 from the National Institutes of Health.

REFERENCES1 Hirai, Y., Takebe, K., Makoto, T., Kobayashi, S. and Takeichi, M. (1992) Cell 69,

471-4812 Montesano, R., Matsumoto, K., Nakamura, T. and Orci, L. (1991) Cell 67,

9091-90983 Plopper, C. G., Hyde, D. M. and Buckpitt, A. R. (1991) The Lung: Scientific

Foundations (Crystal, R. G., West, J. B., Barnes, P. J., Cherniack, J. S. and Weibel,E. R., eds.), vol. 1, pp. 215-228, Raven, New York

4 Massaro, G. D. (1989) Lung Cell Biology (ed. Massaro, D.), pp. 81-114, Dekker,New York

5 McCray, P. B., Wohiford-Lenane, C. L. and Snyder, J. M. (1992) J. Clin. Invest. 90,619-625

6 Gazdar, A. F. and Linnoila, R. I. (1988) Semin. Oncol. 15, 215-2257 Singh, G., Katyal, S. L. and Gottron, S. A. (1985) Biochim. Biophys. Acta 829,

156-1638 Nordlund-Moller, L., Andersson, O., Ahlgren, R., Schilling, J., Giliner, M., Gustafsson,

J. and Lund, J. (1990) J. Biol. Chem. 265, 12690-126939 Hagen, G., Wolf, M., Katayal, S. L., Singh, G., Beato, M. and Sushe, G. (1990)

Nucleic Acids Res. 18, 2939-294610 Wikenseiser, K. A., Wert, S. E., Wispe, J. R., Stahlman, M., D'Amore-Bruno, M.,

Singh, G., Katyal, S. L. and Whitseft, J. A. (1992) Am. J. Physiol. 262, L32-L39

Received 3 March 1993/19 April 1993; accepted 5 May 1993

11 Hackett, B. P., Shimizu, N. and Gitlin, J. D. (1992) Am. J. Physiol. 262, L399-L40612 Stripp, B. R., Sawaya, P. L., Luse, D. S., Wikenheiser, K. A., Wert, S. E., Huffmen,

J. A., Lattier, D. L., Singh, G., Katyal, S. L. and Whitsett, J. A. (1992) J. Biol. Chem.267, 14703-14712

13 Hackett, B. P. and Gitlin, J. D. (1992) Proc. Natl. Acad. Sci. U.S.A. 89, 9079-908314 Voss, J. W. and Rosenfeld, M. G. (1992) Cell 70, 527-53015 Gorski, K., Carneiro, M and Schibler, U. (1986) Cell 47, 767-77616 Bradford, M. M. (1976) Anal. Biochem. 72, 248-25417 Fleming, R. E. and Gitlin, J. D. (1992) J. Biol. Chem. 267, 479-48618 Jones, K. A., Yamamoto, K. R. and Tjian, R. (1985) Cell 42, 559-57219 Costa, R. H., Grayson, D. R. and Darnell, J. E. (1989) Mol. Cell. Biol. 9,1415-142520 Picard, D. and Schaffner, W. (1984) Nature (London) 307, 80-8221 Lai, E., Prezioso, V. R., Tao, W., Chen, W. S. and Darnell, J. E. (1991) Genes Dev. 5,

416-42722 Cao, Z., Umek, R. M. and McKnight, S. L. (1991) Genes Dev. 5,1538-155223 Melton, D. A., Kreig, P. A., Rebagliati, M. R., Maniatis, T., Zinn, K. and Green, M. R.

(1984) Nucleic Acids Res. 12, 7035-705624 Feigner, P. L., Gadek, T. R., Holm, M., Roman, R., Chan, H. W., Wenz, M., Northrop,

J. P., Ringold, G. M. and Danielsen, M. (1987) Proc. Natl. Acad. Sci. U.S.A. 85,7413-7417

25 Weigel, D., Jurgens, G., Kuttner, F., Seifert, E. and Jackle, H. (1989) Cell 57,645-658

26 Grossniklaus, U., Pearson, R. K. and Gehring, W. J. (1992) Genes Dev. 6,1030-1 051

27 Dirksen, M. L. and Jamrich, M. (1992) Genes Dev. 6, 599-60828 Li, C., Lai, C., Sigman, D. S. and Gaynor, R. B. (1991) Proc. Natl. Acad. Sci. U.S.A.

88, 7739-774329 Li, C., Lusis, A. J., Sparkes, R., Tran, S. M. and Gaynor, R. (1992) Genomics 13,

658-66430 Tao, W. and Lai, E. (1992) Neuron 8, 957-96631 Oliver, S. G., van der Aart, J. M., Agostoni-Carbone, M. L., Aigle, M., Alberghina, L.,

Alexandraki, D., Antoine, G., Anwar, R., Ballesta, J. P. G., Benit, P. et al. (1992) Nature(London) 357, 38-46

32 Hacker, V., Grossniklaus, V., Gehring, W. J. and Jackole, H. (1992) Proc. Natl. Acad.Sci. U.S.A. 89, 8754-8758

33 Clevidence, D. E., Overdier, D. G., Tao, W., Qian, X., Pani, L., Lai, E. and Costa, R. H.(1993) Proc. Natl. Acad. Sci. U.S.A. 90, 3948-3952